Coquetting Tarradiddles

Darn Kids, Get Off My Grammar!

Years ago, I couldn’t wait to become an old man so I could sit on my porch, yelling at kids to get off my lawn and griping about politics, taxes, and incontinence to anyone who would listen.

Fortunately, I don’t have to wait until I reach old age to pull out an old chestnut like, “These darn kids today are stupid.” To the casual observer, it might look like I’m just recycling the same tired phrasing that older generations have always applied to newer generations, but there’s more to it than that. Usually, it’s all just so much hot air – a way for older people to feel better about their lives while keeping the whippersnappers in check. Now though, anyone on the other side of the Graduation Day milestone can say it with authority because, quite simply, schools are making their students stupid.

I’m not sure where to place blame, though. I don’t really think it’s anyone’s fault so much as it’s just the natural result of a certain way of thinking that seems to permeate society. There is this underlying faith in “hard data” and that everything – and I do mean everything – can be weighed, measured, quantified and analyzed to pinpoint accuracy, and that the data can then be used to help make things better. It certainly makes sense as a concept, but the problem is that not everything is so easily measured.

For example, a computer can tell you what size an object is. It can even do comparative analysis and tell you that it’s bigger than object x but smaller than object y. What it can’t do is tell you that what you’re looking at is pretty or ugly, or provide any sort of context for consideration. All it can tell you is what some human programmer has told it to say.

If you still don’t believe me, go try an online dating site with some sort of self-professed ‘scientific’ basis for its matchmaking. You put in your interests, your likes and dislikes, and you check off all of the little boxes that define you. If you have a problem defining yourself within the rigid framework of checkboxes, too bad. That’s Science!

Sometimes, the computer will set you up with your perfect match, but it’s just as likely to set you up with your perfect loser. Setting aside for the moment that everyone lies about who they really are, both to others and themselves, it’s hard to imagine any sort of accurate data analysis that computes your perfect match based on multiple-choice answers. What, you like the color red? This guy over here likes the color red, too. Go have babies!

Computers bear a lot of the responsibility for the state of the education system today, and technology advocates are as much to blame as anyone else. People have this feeling that today, with the help of modern computing technology, we can do almost anything. The line of thought goes something like this: Technology improves our lives and helps us reach ever-higher goals, setting new standards today that, through the technology of tomorrow, we can rise above still. The problem with this way of thinking is that it’s based on assumptions that are in turn founded on claims that are dubious at best, and downright fabrications at worst.

People need to understand that computers are deeply stupid things, and that it’s very hard to get them to do anything outside of strict computation. Modern video games and science fiction tend to confuse people into believing that computers are somehow capable of creative thought. They’re not. For all of their complexity, and despite all of the golly-gee whizbangery that we see fly across our television sets, computers remain absolutely and wholly stupid. At their core, computers aren’t really any more complex today than they were thirty or even fifty years ago. When all is said and done, a computer still doesn’t think. It simply computes equations down to a simple binary answer: one or zero, yes or no, on or off.*

And that’s all they do. They don’t synthesize new ideas, they don’t appreciate fine art, and they don’t understand anything. All they do is execute lines of code in a program. They process algorithms designed by a human to achieve a desired result. This ain’t Skynet, kids.*

*2024 update – This is still true in the AI era because Artificial Intelligence Doesn’t Exist, Actually.

Thanks to the general gullibility of the American public, everything started to go wrong in the ’90s with the dot-com boom. Technophiles seemed to be everywhere, preaching the gospel of the internet from their digital pulpits, and people swallowed their nonsense soup, then asked to slurp up seconds.

Today, school districts across America spend ludicrous amounts of money on technology. Every school is wired to the internet via high-speed connectivity. There are multiple computers in every classroom. Entire phone systems have been replaced by sleek, high-tech IP-based telephony. Curriculum departments purchase endless amounts of educational software designed to increase student performance. Teachers and students alike are subjected to identity card swipes, nearly constant camera surveillance, and soon persistent RFID tracking. Schools are great testing grounds for new technology, since tech funding is alarmingly massive, and public acceptance of the spending is almost guaranteed. But what are we getting for our dollars? The short answer is nothing.

The long answer is less than nothing. The law of diminishing returns set in years ago, after multiple computers were placed in every classroom but before we figured out that they’re not really doing any good. (Actually, we still haven’t figured that out. There’s a big push now to put a computer on every student’s desk.*) Computers simply aren’t teachers, and so-called ‘edutainment’ software is light on the ‘edu’ and heavy on the ‘tainment’. So have computers increased student performance?

*Another update from the future: this happened, just with tablets and sometimes laptops.

It’s difficult to answer that question because the whole idea that you can extract any sort of hard data with which to gauge student performance is what has led the school system into the abysmal depths in which it now finds itself. Sure, rating performance is pretty simple with multiple-choice testing, and computers are great at compiling and repackaging that data. The problem is that multiple-choice testing robs the curriculum of any real educational value, and replaces it with rote memorization and a conspicuous absence of critical thought.

Take, for example, a program being sold to schools right now. According to the people doing the selling, it:

“…transforms writing instruction and assessment by applying superior artificial intelligence and linguistic technologies to the writing process. Educators can make timely, data-driven decisions for successful differentiated instruction and motivate students to write more frequently by providing them with immediate feedback.”

That sounds great, doesn’t it? In reality, all the program seems to be is a fancy, much more expensive subscription-based service that provides the same basic function of the “Check Grammar / Readability Statistics” feature of Microsoft Word. Teachers are using this program to instruct students in the art of writing and then grading the results. That’s right. Somehow, a computer that is absolutely incapable of recognizing context and understanding meaning* is now responsible for teaching an entire generation how to write.

*Yet another update from the future: no, ChatGPT can’t do this. LLMs are just really good at pretending they can.

If you’ve ever used the Check Grammar function of Word, you’ll immediately recognize how wrong this is. Assuming that the program never once reported a false error and that it never missed actual errors, and you’re left with a computer system telling a human being how to organize, construct, and relate his or her ideas via the written word – all while having no concept of what words even are.

Try indulging Word whenever it tells you that a sentence needs revision. You’ll find that you soon end up with a very sterile work that not only fails to accurately communicate the ideas you were trying to convey, but that loses any readability it may have once had. This is not how we should be teaching our children to write.

I commit crimes against the language all the time, but I do so for a reason. Usually, it’s for the sake of readability or simply to encourage a particular flow to the language. If the computer had its way, people would never use contractions, would always write in present tense and active voice, and would craft sentences so short in paragraphs so staccato that even Hemingway would cry.

How can anyone with a sound mind think that a computer could ever – or should ever, for that matter – grade a student’s written essay? It has no idea what the sentence you’ve written is about because it doesn’t understand the words. It has no vocabulary, just statistical models of the shape of words it tries to match to patterns in a database and then makes a best guess as to what grammatical rule it thinks you’re breaking.*

*Last update from the future: It’s amazing how well I described how LLMs work 15 years before they existed

Best case, it can grade grammatical mistakes – but writing is so much more than grammar. Teachers use essays to gauge a student’s understanding of the material beyond the simplistic results of multiple-choice testing – but the computer doesn’t understand context, so will a grammatically perfect but contextually nonsensical essay make the grade? Probably. It’s digital snake oil, and school districts are buying it by the bucketful.

At least we can be thankful that such systems haven’t been available throughout history. I can’t imagine reading a version of Huck Finn as written by a computer-assisted Mark Twain, or watching a CPU-corrected rendition of Shakespeare’s works.*

*I guess lied about that last update from the future because, well, here we are. I don’t have to imagine what this would be like anymore. Also, Louisiana is preparing to spend millions on AI tutors instead of hiring more teachers and properly funding public education, which prompted my return to this essay. What a time to be alive.